Thumbnail Image

Using agent-based modelling for small area household projections: a New Zealand case study

In New Zealand, the Resource Management Act 1991 (RMA) is the guiding legislation that sets out how the environment is managed. The territorial authorities are charged with regulating how the environment is managed in their jurisdiction. In order to effectively manage any resource, good and robust planning processes are required. Population change is the factor that has the greatest impact on the environment, and is one of the most challenging to regulate, thus for the territorial authority planners, knowing where and when infrastructure investment is most likely required is key to fulfilling their statutory requirement. This line of investigation has been driven by an ever increasing need for more spatially detailed projections of where and when changes are most likely to take place. The investigation and subsequent development of spatially detailed household projections transcends three territorial authorities in the lower Waikato river catchment situated in New Zealand. An agent-based model was developed in which the individual agents were households. The aim of the model was to produce a simplified simulation of the location choices made by the members of a household in a housing market that is governed by councils’ infrastructure provision, household rents, transport costs and the benefit derived from the neighbourhood amenities and the environment. This model simulates the distribution of households over a twelve year period from 2013 to 2025. In the model the households were programmed to move to vacant properties in order to minimise their residential costs. Each time an agent moves this provides new potential options for all other households; thus the simulation runs and the households move until all households settle in their least cost locations, representing the distribution of households in 2025. The incorporation of multiple territorial authorities provides a more holistic approach than the prevailing approach which is based on disaggregating top level projection with the no further account of population movement outside the top-level migration assumption. The results of the model calibration indicate the model performs well with a 16.6% RMSE at the smallest spatial unit. The projected results produce growth patterns that fall within the expectations of planning staff of the councils. These staff members have affirmed the model’s and input assumptions, and indicate the outcomes to be both useful and plausible. The results demonstrate the relationship between the councils and their respective growth plan strategies. The scenarios were developed to, demonstrate the relationship between the availability of vacant land and the cost to occupy a property, which ultimately impacts the flux of residents to or away from the city. With some broad assumptions and limitations, this model is distinctive in its approach as it is developed with the intention of being an applied tool to be used by the three councils. It has the further distinction as modelling the behaviours of individual households rather than the behaviours of an entire population, which is the unique in New Zealand and amongst a few applied models of this nature wide (Triantakonstantis & Mountrakis, 2012).
Type of thesis
Marais, A. L. (2016). Using agent-based modelling for small area household projections: a New Zealand case study (Thesis, Master of Social Sciences (MSocSc)). University of Waikato, Hamilton, New Zealand. Retrieved from https://hdl.handle.net/10289/10589
University of Waikato
All items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.